logsite

Your logs say something broke. logsite tells you which line of your code, and writes a test that proves it.

It reads the logs you already have (a file, Loki, OpenSearch, CloudWatch, Sentry or OpenTelemetry), groups thousands of stack traces into a few causes, and points each one at your line of code. No SDK, nothing to add to your services. Works with Python, Java, Go, TypeScript and JavaScript.

pip install logsite

Try it

The repo has a small demo shop with a planted bug and its production log.

pip install logsite pytest
git clone https://github.com/rajpaynaik/logsite && cd logsite/examples/bookstore

1. What's failing?

$ logsite errors --from ../../bench/repro/scenarios/supplier-timeout/incident.log
28 errors in incident.log · 2 group(s)

  #  TREND  COUNT  ERROR                                      RAISED AT       LOGGED AT
  1            25  TimeoutError: timed out                    supplier.py:31  api.py:36
  2             3  ConnectionRefusedError: cache unavailable  cache.py:15     api.py:23

2. Write a test that reproduces it.

$ logsite repro --from ../../bench/repro/scenarios/supplier-timeout/incident.log --write-test
wrote tests/test_repro_346fa937.py

3. Check it fails the way production did.

$ logsite agent --expect 346fa937 -- python -m pytest tests/test_repro_346fa937.py
✓ 346fa937 reproduced: TimeoutError through bookstore/supplier.py:31, as in production

Fix supplier.py until the test passes.

Use it on your service

cd your-service
logsite errors --from app.log        # errors grouped by cause
logsite lint                         # logging mistakes in your code

To read from your log store, add a logsite.toml:

[backend]
type = "loki"                        # or "opensearch", "cloudwatch", "sentry", "file"
url = "http://localhost:3100"
selector = '{service_name="checkout"}'

Watch production

logsite watch checks your logs every 5 minutes, stores every error in your own database, and posts to Slack when an error is new, back after a fix, or spiking. No SDK: it reads the log store you already have.

In the browser: logsite app walks you through the log store, Slack and database, each with a test button, then starts watching.

Or with a config file:

[backend]
type = "loki"
url = "http://loki:3100"
selector = '{service_name="checkout"}'

[alerts]
slack_webhook_env = "SLACK_WEBHOOK_URL"   # a Slack incoming webhook

[store]                                   # leave out to use SQLite
url_env = "LOGSITE_DATABASE_URL"          # Postgres, MySQL, SQL Server or ClickHouse

pip install "logsite[postgres]"
logsite watch --dry-run                   # print the Slack messages first
logsite watch

Tables are created on first use. The first check loads the last 24 hours without posting; logsite db backfill --since 30d loads more.

Where to run it: anywhere that can reach your log store: a server, Docker, or Kubernetes. On Kubernetes (EKS included) it's one Deployment with one replica, config in a ConfigMap, secrets in a Secret, and an IAM role for CloudWatch. See Run on Kubernetes.

More

Also: a GitHub Action, a VS Code extension, and a Claude connector (MCP) so you can ask "what failed in checkout yesterday after 3pm?" in plain English.

Docs

Commands · Configuration · logsite app · Watch production · Kubernetes · Reproduce an error · AI prompt · Security · Changelog

Questions and ideas: Discussions. Contributions welcome, see CONTRIBUTING.md.

License

Apache 2.0